Passive Voice (Typology)

What the passive does

Take a transitive sentence. It has two participants: an agent (the doer) and a patient (the thing done to). In the active voice, English maps these to subject and object:

The dog       bit       the postman.
agent/SUBJ    verb      patient/OBJ

The passive promotes the patient to subject and demotes the agent. The agent can be expressed in an oblique phrase (the by-phrase) or simply dropped:

The postman   was bitten   (by the dog).
patient/SUBJ  passive verb   agent/OBL (optional)

Crucially, the event is the same — a single biting. What changes is informational packaging. The passive lets the speaker put the patient in topic position (where new sentences usually pick up old information) and leave the agent unsaid when it's irrelevant, unknown, generic, or politically inconvenient. "Mistakes were made."

Three things are happening simultaneously: (1) argument promotion — patient takes the subject slot; (2) argument demotion — agent loses the subject slot, becomes optional; (3) verbal marking — the verb signals that this rearrangement has occurred. Different languages do all three with very different machinery.

English: periphrastic be + past participle

English uses two words: an auxiliary (typically be, sometimes get) plus the past participle of the main verb.

Active:   Mary  wrote     the letter.
Passive:  The letter  was  written  (by Mary).

The participle written is the same form used in the perfect (has written), which can confuse learners. The disambiguator is the auxiliary: has + past-participle is perfect; be + past-participle is passive. The agent appears in a by-phrase, marked with the preposition by.

English passives are productive — almost any transitive verb can be passivized — but ditransitives offer two options: "John gave Mary the book" passivizes as either "Mary was given the book" (recipient promoted) or "The book was given to Mary" (theme promoted). Other languages typically allow only one.

Japanese: morphological -rare- and the indirect passive

Japanese marks passive with the verbal suffix -(r)are-, attached directly to the verb stem. The agent is marked with the postposition ni (or sometimes kara):

Active:   犬が郵便配達員を噛んだ。
          inu-ga  yuubin-haitatsu-in-o    kanda
          dog-NOM postman-ACC             bit
          'The dog bit the postman.'

Passive:  郵便配達員が犬に噛まれた。
          yuubin-haitatsu-in-ga  inu-ni  kam-are-ta
          postman-NOM            dog-DAT bit-PASS-PAST
          'The postman was bitten by the dog.'

So far this is similar to English. But Japanese has a second passive — the indirect passive (also called the adversative passive) — that has no English equivalent. It applies to intransitive verbs and signals that the subject was adversely affected by an event:

太郎が雨に降られた。
Tarou-ga  ame-ni  fur-are-ta
Taro-NOM  rain-DAT fall-PASS-PAST
'Taro got rained on' / 'Taro was adversely affected by rain falling.'

The verb furu "fall" is intransitive — there is no patient to promote. But Japanese still allows passivization, with the meaning that the subject suffers from the event. The passive isn't just argument-rearrangement; it carries a flavor of adverse experience.

Mandarin: the bei-construction

Mandarin uses an analytic strategy — a separate word, bei (被), placed before the agent:

Active:   狗咬了郵差。
          gǒu  yǎo-le      yóuchāi
          dog  bit-PERF    postman
          'The dog bit the postman.'

Passive:  郵差被狗咬了。
          yóuchāi  bèi  gǒu  yǎo-le
          postman  BEI  dog  bit-PERF
          'The postman was bitten by the dog.'

Classical Mandarin bei carried strong adversative semantics — the subject suffers. "我被欺騙了" (wǒ bèi qīpiàn-le, "I was deceived") is fine; "我被誇獎了" (wǒ bèi kuājiǎng-le, "I was praised") was traditionally awkward. Modern usage, under European-language influence, accepts neutral and even positive bei-passives, but the adversative undertone lingers.

Mandarin also has agent-less bei: "信被寄了" (xìn bèi jì-le, "the letter was sent"). And it has a longer construction bei + agent + gěi + verb that some grammarians treat as a separate passive type. The system is grammatically simpler than English (no participle morphology) but semantically richer (adversative meaning baked in).

Tagalog: the symmetrical voice system

Tagalog and most Philippine languages have something quite different — sometimes called symmetrical voice, the Philippine-type voice system, or historically just "focus." Instead of an active/passive contrast, there are four (or more) voices, all morphologically marked, none clearly basic:

Actor voice:        Bumili    ang  lalaki  ng    isda  sa  tindahan.
                    AV.bought TOP  man     OBL   fish  LOC store
                    'The man bought a fish at the store.'

Patient voice:      Binili    ng    lalaki  ang  isda   sa  tindahan.
                    PV.bought OBL   man     TOP  fish   LOC store
                    'The man bought THE FISH at the store.'

Locative voice:     Binilihan ng    lalaki  ng   isda  ang  tindahan.
                    LV.bought OBL   man     OBL  fish  TOP  store
                    'The man bought a fish at THE STORE.'

Benefactive voice:  Ibinili   ng    lalaki  ng   isda  ang  bata  sa tindahan.
                    BV.bought OBL   man     OBL  fish  TOP  child LOC store
                    'The man bought a fish FOR THE CHILD at the store.'

Each voice puts a different argument in the privileged slot marked by ang (the topic/pivot). Crucially, none is the "default" — all four are equally productive, equally common, and equally morphologically marked. This is why most typologists no longer call the patient voice a passive: passives by definition are marked alternatives to an unmarked active, and Tagalog has no unmarked baseline.

Cross-linguistic comparison

	English	Japanese	Mandarin	Tagalog	German	Latin
Marking type	Periphrastic (be + ptcp)	Morphological (-rare-)	Analytic (bei)	Affixation (-in-/-an/i-)	Periphrastic (werden/sein)	Morphological (-r endings)
Agent phrase	by-phrase	ni-phrase	bei NP	ng-phrase (oblique)	von / durch	ablative + ā/ab
Adversative semantics	No	Yes (indirect passive)	Yes (classical)	No	No	No
Passive of intransitive	No	Yes (indirect)	No	N/A	Yes (impersonal)	Yes (impersonal)
Stative vs dynamic split	Context-dependent	Context-dependent	Context-dependent	N/A	sein vs werden	Context-dependent
Status of "active"	Default, unmarked	Default, unmarked	Default, unmarked	One of several voices	Default, unmarked	Default, unmarked
Productivity	High (most transitives)	High	Lower (favors adversative)	All voices fully productive	High	High

The table flattens nuance — every language has restrictions and idiosyncrasies — but the patterns are real. English and German share a periphrastic strategy; Japanese and Latin share a morphological one; Mandarin alone has an analytic dedicated word; Tagalog stands apart entirely.

Worked example: passivizing one event in five languages

Consider the event: a child opened a door. Here's how each language packages it actively, then passively:

Language	Active	Passive
English	The child opened the door.	The door was opened (by the child).
Japanese	子供がドアを開けた。 kodomo-ga doa-o ake-ta	ドアが子供に開けられた。 doa-ga kodomo-ni ake-rare-ta
Mandarin	小孩打开了门。 xiǎohái dǎkāi-le mén	门被小孩打开了。 mén bèi xiǎohái dǎkāi-le
Tagalog (PV)	Binuksan ng bata ang pinto. (patient = ang pinto)	(no separate passive — patient voice is the same form)
German	Das Kind öffnete die Tür.	Die Tür wurde (vom Kind) geöffnet. (dynamic) Die Tür ist geöffnet. (stative)

Notice that Tagalog has nothing extra to do — the "passive-like" reading is just one of the basic voice options, available without any added morphology. German has two passives and forces you to commit to dynamic vs stative. Mandarin's bei works fine here because opening a door is neutral, but would feel marked for joyful events.

Voice variants and relatives

• Middle voice. Found in Greek, Sanskrit, Latin, Romance reflexives. The subject is both agent and affected — "the door opened" (English uses an unaccusative; Spanish uses se: la puerta se abrió). Middle voice blurs into passive in some languages.
• Antipassive. The mirror of passive in ergative-absolutive languages: the agent becomes the prominent argument and the patient is demoted. Common in Mayan, Inuit, Australian languages.
• Impersonal passive. No promoted subject; agent suppressed. German "Es wurde getanzt" (literally "It was danced" = "There was dancing"). Latin and Polish allow this freely.
• Get-passive. English alternative: "He got fired." Often more colloquial, often with adversative or unintended-event flavor — "He was fired" is neutral, "He got fired" suggests bad luck or initiative.
• Adjectival passive. Stative: "The door is closed." Behaves like an adjective (gradable: "very closed"; with become: "became closed"). Distinct from verbal passive in syntactic tests.
• Pseudo-passive (prepositional). English allows "The bed was slept in" — the prepositional object is promoted. Most languages don't allow this.
• Long vs short passive. Long = with by-phrase ("by Mary"); short = without. Across languages, short passives are dramatically more frequent than long ones — agent omission is the main pragmatic motivation.

Why passive typology matters

• Universal vs language-specific. The passive is a probe for what UG-style universals look like — patient promotion appears widely, but the marking and restrictions vary, suggesting the universal is functional, not formal.
• Information structure. Passives are a primary way languages let speakers control topic and given/new. Studying them clarifies what discourse structure looks like cross-linguistically.
• Acquisition order. Children acquire active before passive in every language tested, but the gap varies — English children master full passives around age 5; Sesotho children much earlier, because passives are far more frequent in their input.
• Aphasia and processing. Broca's aphasics struggle disproportionately with reversible passives ("the boy was hit by the girl") — a classic finding that supports syntactic-deficit theories of agrammatism.
• Translation and NLP. Mismatched passive systems are a major source of stilted machine translation. Translating Mandarin bei-passives into English without their adversative tone loses meaning.
• Argument-structure theory. Passives are the workhorse construction for testing claims about how syntactic structures encode thematic roles.

Common pitfalls

• Conflating passive with stative. "The door was closed" can be either; tests with manner adverbs or by-phrases distinguish them. German morphologizes the distinction (sein vs werden); English forces context to do the work.
• Assuming Mandarin bei = English by. Mandarin bei is a verb-like marker introducing the agent and carries adversative flavor; English by is a neutral preposition. Direct translation often sounds wrong.
• Calling Tagalog patient voice "passive." Older descriptive grammars do this — but Tagalog has no unmarked active, so the "passive" label misleads. Use "patient voice" or "non-actor voice."
• Overusing passive in style. The passive is a tool, not a sin. Strunk-and-White scolding misses the point: passives are perfect when the agent is irrelevant or unknown. Forcing actives into agentless contexts often reads as awkward — "Someone heated the samples" is worse than "The samples were heated" in a methods section.
• Treating "get" and "be" passives as equivalent. "He was fired" and "He got fired" are both passive in form, but differ in flavor — get-passives often imply unintended or unwelcome events, similar to Mandarin bei.
• Forgetting impersonal passives exist. German and Polish allow passivizing intransitives with no patient at all. The passive is fundamentally about agent demotion; patient promotion is a side effect when there is one.